---
 libavresample/x86/audio_convert.asm    |   45 ++++++++++++++++++++++++++++++++
 libavresample/x86/audio_convert_init.c |    5 +++
 2 files changed, 50 insertions(+), 0 deletions(-)

diff --git a/libavresample/x86/audio_convert.asm 
b/libavresample/x86/audio_convert.asm
index 8159188..a706b17 100644
--- a/libavresample/x86/audio_convert.asm
+++ b/libavresample/x86/audio_convert.asm
@@ -316,6 +316,51 @@ cglobal conv_s16p_to_s16_6ch, 2,8,6, dst, src, src1, src2, 
src3, src4, src5, len
     jg .loop
     REP_RET
 
+;------------------------------------------------------------------------------
+; void ff_conv_s16p_to_flt_2ch(float *dst, int16_t *const *src, int len,
+;                              int channels);
+;------------------------------------------------------------------------------
+
+INIT_XMM sse2
+cglobal conv_s16p_to_flt_2ch, 3,4,6, dst, src0, len, src1
+    lea       lenq, [2*lend]
+    mov      src1q, [src0q+gprsize]
+    mov      src0q, [src0q        ]
+    lea       dstq, [dstq+4*lenq]
+    add      src0q, lenq
+    add      src1q, lenq
+    neg       lenq
+    mova        m5, [pf_s32_inv_scale]
+    ALIGN 16
+.loop:
+    mova        m2, [src0q+lenq]
+    mova        m3, [src1q+lenq]
+    punpcklwd   m0, m2, m3
+    punpckhwd   m1, m2, m3
+    pxor        m2, m2
+    pxor        m3, m3
+    punpcklwd   m2, m0
+    punpckhwd   m3, m0
+    pxor        m0, m0
+    pxor        m4, m4
+    punpcklwd   m0, m1
+    punpckhwd   m4, m1
+    cvtdq2ps    m2, m2
+    cvtdq2ps    m3, m3
+    cvtdq2ps    m0, m0
+    cvtdq2ps    m4, m4
+    mulps       m2, m5
+    mulps       m3, m5
+    mulps       m0, m5
+    mulps       m4, m5
+    mova  [dstq+4*lenq         ], m2
+    mova  [dstq+4*lenq+  mmsize], m3
+    mova  [dstq+4*lenq+2*mmsize], m0
+    mova  [dstq+4*lenq+3*mmsize], m4
+    add       lenq, mmsize
+    js .loop
+    REP_RET
+
 ;-----------------------------------------------------------------------------
 ; void ff_conv_fltp_to_flt_6ch(float *dst, float *const *src, int len,
 ;                              int channels);
diff --git a/libavresample/x86/audio_convert_init.c 
b/libavresample/x86/audio_convert_init.c
index e4708de..93e2b99 100644
--- a/libavresample/x86/audio_convert_init.c
+++ b/libavresample/x86/audio_convert_init.c
@@ -48,6 +48,9 @@ extern void ff_conv_s16p_to_s16_2ch_sse2(int16_t *dst, 
int16_t *const *src,
 extern void ff_conv_s16p_to_s16_6ch_sse2(int16_t *dst, int16_t *const *src,
                                          int len, int channels);
 
+extern void ff_conv_s16p_to_flt_2ch_sse2(float *dst, int16_t *const *src,
+                                         int len, int channels);
+
 extern void ff_conv_fltp_to_flt_6ch_mmx (float *dst, float *const *src, int 
len,
                                          int channels);
 extern void ff_conv_fltp_to_flt_6ch_sse4(float *dst, float *const *src, int 
len,
@@ -85,6 +88,8 @@ av_cold void ff_audio_convert_init_x86(AudioConvert *ac)
                                   2, 16, 16, "SSE2", 
ff_conv_s16p_to_s16_2ch_sse2);
         ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_S16P,
                                   6, 16, 8, "SSE2", 
ff_conv_s16p_to_s16_6ch_sse2);
+        ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16P,
+                                  2, 16, 8, "SSE2", 
ff_conv_s16p_to_flt_2ch_sse2);
     }
     if (mm_flags & AV_CPU_FLAG_SSE4 && HAVE_SSE) {
         ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_S16,
-- 
1.7.1

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to